Sun et al. Diabetol Metab Syndr (2019) 11:51 https://doi.org/10.1186/s13098-019-0450-x Diabetology &

RESEARCH Open Access Assessment of adiposity distribution and its association with and insulin resistance: a population‑based study Kan Sun, Diaozhu Lin, Qiling Feng, Feng Li, Yiqin Qi, Wanting Feng, Chuan Yang, Li Yan, Meng Ren and Dan Liu*

Abstract Background: Rational measures in estimating adiposity distribution in diabetic patients has yet to be validated. This study aims to provide insight about the possible links between routinely available body adiposity parameters and the development of both diabetes and insulin resistance. Methods: We performed a population-based cross-sectional study in 9496 subjects aged 40 years or older. All of the body adiposity measures including (BMI), waist circumference (WC), waist-hip ratio (WHR), waist- height ratio (WHtR), visceral adiposity index (VAI), body adiposity index (BAI) and lipid accumulation product index (LAP) were separately evaluated according to standard measurement methods. Diabetes was diagnosed according to the American Diabetes Association 2010 criteria. Results: All tested body adiposity measurements were signifcantly associated with fasting plasma glucose (FPG), oral glucose tolerance test (OGTT) 2 h glucose, HbA1c and fasting insulin. Compared with other adiposity pheno- types, LAP have shown the relatively strongest while BAI have shown the relatively weakest association with increased odds of both diabetes and insulin resistance across all logistic regression models. Additionally, LAP provided the best discrimination accuracy for diabetes [area under the curve (AUC): 0.658 95% confdence intervals (CI) 0.645–0.671] and insulin resistance (AUC: 0.781 95% CI 0.771–0.792) when compared with other body adiposity parameters. Conclusions: The LAP index seems to be a better indicator than other adiposity measures tested in the study to evaluate the association of visceral fat mass with diabetes and insulin resistance, which should be given more consid- eration in the clinical practice. Keywords: Diabetes, Insulin resistance, Body adiposity parameters, Visceral adiposity index, Body adiposity index, Lipid accumulation product index

Introduction anthropometric measurements and surrogate adipose Diabetes is a major risk factor for premature death in the indices are used in feld settings to assess the risk of dia- general population and results in a huge social and eco- betes and insulin resistance [4, 5]. However, the rational nomic burden on the health systems worldwide [1, 2]. and preferred method in estimating body adiposity in is closely associated with a higher incidence of diabetes patients has yet to be validated. and involved in the development of dia- Body mass index (BMI) is one of the most commonly betic complications [3]. Considering the important char- used clinical measure to determine obesity in individu- acteristic of excessive body fat accumulation, numerous als. However, when using as a measurement of visceral and subcutaneous body fat distribution, the limitations of BMI is also well documented in various populations *Correspondence: [email protected] [6–8]. When comparing estimates of anthropometric Department of Endocrinology, Sun Yat‑sen Memorial Hospital, Sun measures with respect to their ability to predict the per- Yat-sen University, 107 Yanjiang West Road, Guangzhou 510120, People’s Republic of China centage of body fat, Schulze et al. [9] found that waist

© The Author(s) 2019. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/ publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Sun et al. Diabetol Metab Syndr (2019) 11:51 Page 2 of 10

circumference (WC) in men was considerably better cor- Subjects who failed to provide information [WC: n = 186; related with body fat than BMI. hip circumference (HC): n = 42; height: n = 29; weight: Dual-energy X-ray absorptiometry (DXA) is consider n = 23; triglycerides (TG): n = 23; high-density lipopro- as one of the best methods in quantifying individual tein cholesterol (HDL-C): n = 2; fasting plasma glucose adiposity. Recently, body adiposity index (BAI) was pro- (FPG): n = 15; oral glucose tolerance test (OGTT) 2 h posed as a useful measure for quantifying adiposity of glucose: n = 60; HbA1c: n = 40] were excluded from the individuals, which was validated with strong accuracy analyses. Accordingly, a total of 9496 eligible individuals and correlation with DXA-derived body fat composition were included in the fnal data analyses. Te study pro- [10]. However, clinical utility of body composition using tocol was approved by the Institutional Review Board BAI is criticized in later confrmation studies as the rela- of the Sun Yat-sen Memorial Hospital afliated to Sun tive accuracy of BAI in predicting percentage of body fat Yat-sen University and was in accordance with the prin- is strongly confounded by sex and the degree of obesity ciple of the Helsinki Declaration II. Written informed [11, 12]. consent was obtained from each participant before data Te visceral adiposity index (VAI) is a valuable indi- collection. cator of visceral adiposity and dysfunc- tion [13]. Research in various populations reported that Clinical and biochemical data collection VAI is closely associated with many health-related out- Information on lifestyle factors, sociodemographic char- comes such as insulin resistance and type 2 diabetes acteristics and family history were collected by using a [14, 15]. However, in obese elderly women, variation of standard questionnaire. Smoking or drinking habit was BAI, but not VAI, was positively associated with diabetes classifed as ‘never’, ‘current’ (smoking or drinking regu- related infammatory cytokines including tumor necrosis larly in the past 6 months) or ‘ever’ (cessation of smoking factor-α and interleukin-6 [16, 17]. or drinking more than 6 months) [25]. A short form of As a new anthropometric measure of lipid over accu- the International Physical Activity Questionnaire (IPAQ) mulation, the accurate predictive value of lipid accumula- was used to estimate physical activity at leisure time by tion product index (LAP) in insulin sensitivity, diabetes adding questions on frequency and duration of moderate and metabolic syndrome is also documented [18–20]. In or vigorous activities and walking [26]. Separate meta- addition, LAP was strongly correlated with interleukin-6 bolic equivalent hours per week (MET-h/week) were cal- levels and many other adipocytokines based on recent culated for evaluation of total physical activity. publications [21]. Venous blood samples were collected for laboratory Distribution of body fat accumulation is of great tests after an overnight fasting of at least 10 h. Measure- important in the development of insulin resistance and ments of FPG, fasting serum insulin, TG, total choles- diabetes. However, the inconclusive relationship of body terol (TC), HDL-C, low-density lipoprotein cholesterol adipose composition with diabetes might have negative (LDL-C), creatinine and γ-glutamyltransferase (γ-GGT) efect when determining the prevention and treatment were done using an autoanalyser (Beckman CX-7 Bio- strategies for the disease. To date, previously existing chemical Autoanalyser, Brea, CA, USA). HbA1c was investigations have not been implemented to make a assessed by high-performance liquid chromatography systematic comparison on the associations of all these (BioRad, Hercules, CA). Te abbreviated Modifcation of adipose indices with diabetes and insulin resistance. We in Renal Disease (MDRD) formula recalibrated for have therefore aimed to determine a relatively superior Chinese population was used to calculate estimated glo- body adiposity parameter associated with both diabetes merular fltration rate (eGFR) expressed in mL/min per 2 and insulin resistance in a community-based population. 1.73 m using a formula of eGFR = 175 × [serum creati- −1.234 −0.179 nine × 0.011] × [age] × [0.79 if female], where Subjects and methods serum creatinine was expressed as μmol/L [27]. Study population and design We performed a cross-sectional study in a community Anthropometric and body adiposity measurements in Guangzhou, China from June to November 2011. Te All participants completed anthropometrical meas- study population was from the REACTION study and urements with the assistance of trained staf by using details of this study have been published previously [22– standard protocols. Body height and body weight were 24]. At frst, 10,104 subjects aged 40 years or older were recorded to the nearest 0.1 cm and 0.1 kg while partici- invited to participate by examination notices or home pants were wearing light indoor clothing without shoes. visits during the recruiting phase. After this, 9916 sub- BMI was calculated as weight in kilograms divided by jects signed the consent form and agreed to participate height in meters squared (kg/m2). WC was measured at in the survey, including those with or without diabetes. the umbilical level with participant in standing position, Sun et al. Diabetol Metab Syndr (2019) 11:51 Page 3 of 10

at the end of gentle expiration. HC was measured the diabetes. Linear regression analyses were used to test same way as WC, but the measuring tape was placed for trends across groups. Pearson’s correlation and around the widest part of the buttocks. Waist-to-hip ratio multiple regression analysis adjusted for age and sex (WHR) was calculated as WC divided by HC while waist- were performed to test the correlations of body adipos- to-height ratio (WHtR) was calculated as WC divided ity parameters with FPG, OGTT 2 h glucose, HbA1c by height in cm. VAI was determined by gender-specifc and fasting insulin. Te unadjusted and multivari- equations and calculated using the following formulas. ate-adjusted logistic regression analysis was used to Men: [WC/(39.68 + (1.88 × BMI))] × (TG/1.03) × (1.31/ assess the risk of increased prevalent insulin resist- HDL-C); Women: [WC/(36.58 + (1.89 × BMI))] × (TG ance and diabetes in relation to 1-quartile increase in /0.81) × (1.52/HDL-C) [28]. WC was calculated in cm body adiposity measurements. Model 1 is unadjusted. while HDL-C and TG in mmol/L. BAI was calculated Model 2 is adjusted for age. Model 3 is adjusted for 1.5 using the equation as [(HC/(height ) − 18] [10]. HC age, sex, current smoking and drinking status, physi- and height were calculated in cm. LAP was calculated as cal activity level, systolic blood pressure (SBP), LDL-C, (WC − 65) × TG in men and (WC − 58) × TG in women. γ-GGT, eGFR and antidiabetic treatment. By consid- WC was calculated in cm and TG in mmol/L [29, 30]. ering the variation in each quartile group, multivar- Tree times consecutively blood pressure measurements iate-adjusted logistic regression analysis was used to were obtained by an automated electronic device in a assess the prevalent diabetes and insulin resistance 5-min interval by the same observer (OMRON, Omron in relation to each quartile increase in body adiposity Company, China). Te average of three measurements of measurements. Odds ratios (OR) and the correspond- blood pressure was used for analysis. ing 95% confdence intervals (CI) were calculated. Te discriminative ability of adiposity measurements on prevalent insulin resistance and diabetes was examined Defnition of insulin resistance and diabetes through area under the receiver operating character- Te insulin resistance index (homeostasis model assess- istic curve (AUC) with 95% CI. Diferences between ment of insulin resistance, HOMA-IR) was calculated as the AUC of anthropometric measures were performed with a nonparametric approach [34]. To obtain a bet- fasting insulin (μIU/mL) × FPG (mmol/L)/22.5 [31]. Insu- lin resistance was defned by a HOMA-IR index within ter assessment of the estimation power of the adipose the top quartile (greater than 2.54) in the present study distribution parameters, we used the optimal operating [32]. Diabetes was diagnosed according to the American point with setting the minimum sensitivity of 70% for Diabetes Association 2010 criteria, which included (1) which we do not want sensitivity to fall below [35]. FPG level of 7.0 mmol/L or greater, or (2) OGTT 2 h glu- All statistical tests were two-sided, and a P cose level of 11.1 mmol/L or greater, or (3) HbA1c level value < 0.05 was considered statistically signifcant. of 6.5% or greater [33]. Results Statistical analysis Basic characteristics of the study population Statistical analysis was performed using SAS version 9.2 Te mean age was 55.9 ± 8.1 years among the 9496 (SAS Institute Inc, Cary, NC, USA). Continuous vari- enrolled individuals. Totally, 2054 subjects diagnosed with diabetes and the prevalence rate was 21.6% in ables were presented as mean ± standard deviation (SD) except for skewed variables, which were presented as this population. Accordingly, 606 (6.4%) subjects have medians (interquartile ranges). Categorical variables been diagnosed with diabetes before the survey and were expressed as numbers (proportions). Diferences among them 479 (5.0%) subjects were receiving anti- between groups were performed by one-way ANOVAs. diabetic treatment. Moreover, there were 1448 (16.3%) Comparisons between categorical variables were per- newly diagnosed diabetic patients based on the survey. formed by the χ2 test. Physical activity levels, TG, FPG, Te demographic and biochemical characteristics of OGTT 2 h glucose, HbA1c, fasting insulin, HOMA-IR, the study population were shown in Table 1. Subjects γ-GGT, WHR, WHtR, VAI, BAI and LAP were logarith- in diabetes group difered from those in non-diabetes mically transformed before analysis due to a non-normal group in multiple clinical variables. All of the adiposity distribution. measurements in the study were signifcantly elevated We primarily analyzed the associations of body fat in diabetes group (all P < 0.0001). Moreover, LAP was content estimates (BMI, WC, WHR, WHtR, VAI, BAI dramatically elevated in participants with diabetes in and LAP) with FPG, OGTT 2 h glucose, HbA1c, fast- comparison with those without the disease [24.4 (14.4– ing insulin and prevalence of insulin resistance and 41.2) vs 38.9 (22.5–61.9), P < 0.0001]. Sun et al. Diabetol Metab Syndr (2019) 11:51 Page 4 of 10

Table 1 Characteristics of study population correlation coefcient with FPG (r = 0.22, P < 0.0001), Non-diabetes Diabetes P value OGTT 2 h glucose (r = 0.27, P < 0.0001), HbA1c (r = 0.24, P < 0.0001) and fasting insulin (r 0.52, P < 0.0001). In n (%) 7442 (78.4) 2054 (21.6) < 0.0001 = multivariate linear regression analysis adjusted for age Age (years) 55.1 7.7 58.8 8.6 < 0.0001 ± ± and sex, LAP was considerably most strongly correlated Male [n (%)] 2062 (27.7) 630 (30.7) 0.008 with HbA1c (β 0.22, P < 0.0001) and fasting insulin Physical activity (MET-h/ 22.4 (10.5–45.0) 21.0 (10.5–42.0) 0.497 = week) (β = 0.52, P < 0.0001) while VAI was most strongly corre- Current smoking [n (%)] 731 (10.0) 204 (10.2) 0.817 lated with FPG (β = 0.21, P < 0.0001) and OGTT 2 h glu- Current drinking [n (%)] 238 (3.3) 75 (3.7) 0.311 cose (β = 0.26, P < 0.0001). However, as an independent SBP (mmHg) 124.4 16.1 132.2 16.8 < 0.0001 determinant of body fat percentage, BAI was most weakly ± ± DBP (mmHg) 74.9 9.8 76.9 9.9 < 0.0001 associated with FPG, OGTT 2 h glucose and HbA1c than ± ± TG (mmol/L) 1.21 (0.89–1.73) 1.57 (1.08–2.26) < 0.0001 other body adiposity measurements in both Pearson’s TC (mmol/L) 5.17 1.22 5.28 1.36 0.0006 correlation and multivariate linear regression analysis. As ± ± HDL-C (mmol/L) 1.34 0.36 1.23 0.35 < 0.0001 shown in Fig. 1, with the elevating quartiles of adiposity ± ± LDL-C (mmol/L) 3.13 0.94 3.18 1.03 0.040 phenotypes in the study, levels of FPG, OGTT 2 h glu- ± ± cose and HbA1c were all signifcantly increased (all P for FPG (mmol/L) 5.29 (4.93–5.67) 6.46 (5.71–7.69) < 0.0001 trend < 0.0001). OGTT 2 h glucose 6.90 (5.87–8.10) 11.67 (9.39–13.55) < 0.0001 (mmol/L) HbA1c (%) 5.80 (5.60–6.10) 6.60 (6.30–7.20) < 0.0001 Body adiposity parameters with diabetes and insulin Fasting insulin (μIU/mL) 6.80 (5.10–9.40) 8.70 (6.00–12.40) < 0.0001 resistance HOMA-IR 1.60 (1.16–2.25) 2.61 (1.75–3.85) < 0.0001 After this, we further studied the prevalence of diabetes γ-GGT (U/L) 19.0 (14.0–27.0) 24.0 (17.0–36.0) < 0.0001 and insulin resistance in diferent quartiles of body adi- eGFR (mL/min per 102.3 22.6 99.5 27.1 < 0.0001 posity measurements. As showed in Fig. 2, the change in 2 ± ± 1.73 m ) proportion of prevalent diabetes and insulin resistance BMI 23.4 3.3 24.7 3.5 < 0.0001 ± ± were increased with the elevated quartiles of all tested WC (cm) 80.7 9.2 85.2 9.9 < 0.0001 ± ± adiposity parameters in the study (all P for trend < 0.001). WHR 0.86 (0.82–0.90) 0.89 (0.85–0.94) < 0.0001 Te positive associations of BMI, WC, WHR, WHtR, WHtR 0.51 (0.47–0.54) 0.54 (0.50–0.58) < 0.0001 VAI, BAI and LAP with diabetes and insulin resistance VAI 1.55 (1.03–2.44) 2.20 (1.40–3.50) < 0.0001 were consistently detected in both univariable and mul- BAI 28.9 (26.3–31.7) 29.7 (26.8–32.9) < 0.0001 tivariable logistic regression analysis (Tables 3, 4). As LAP 24.4 (14.4–41.2) 38.9 (22.5–61.9) < 0.0001 shown in Table 4, compared with participants in quartile Data were mean SD or medians (interquartile ranges) for skewed variables or 1 of body adiposity measurements, multivariate-adjusted ± numbers (proportions) for categorical variables logistic regression analysis showed that participants in 2 P values were for the ANOVA or χ analyses across the groups quartile 2, quartile 3 and quartile 4, respectively, have a MET-h/wk separate metabolic equivalent hours per week, SBP systolic blood pressure, DBP diastolic blood pressure, TG triglycerides, TC total cholesterol, signifcant correlation with increased odds of diabetes HDL-C high-density lipoprotein cholesterol, LDL-C low-density lipoprotein and insulin resistance. Compared with other adiposity cholesterol, FPG fasting plasma glucose, OGTT​ oral glucose tolerance test, phenotypes in the study, LAP have shown the relatively HOMA-IR homeostasis model assessment of insulin resistance, eGFR estimated glomerular fltration rate, γ-GGT​ γ-glutamyltransferase, BMI body mass index, WC strongest while BAI have shown the relatively weakest waist circumference, WHR waist-hip ratio, WHtR waist-height ratio, VAI visceral associations with increased odds of both diabetes and adiposity index, BAI body adiposity index, LAP lipid accumulation product index insulin resistance across all logistic regression models. * P < 0.05 compared with non-diabetes group Further receiver operating characteristics curve repre- sented consistent results and the LAP still performed the Body adiposity parameters and metabolic factors best discrimination accuracy for diabetes (AUC: 0.658 of glycometabolism 95% CI 0.645–0.671, all P < 0.05 except WHR) and insulin Pearson’s correlation analysis revealed that body adipos- resistance (AUC: 0.781 95% CI 0.771–0.792, all P < 0.05) ity measurements including BMI, WC, WHR, WHtR, when compared with other adiposity phenotypes. We VAI, BAI and LAP were all signifcantly associated with analysis the optimal operating point for the screening increased FPG, OGTT 2 h glucose, HbA1c and fasting diabetes of the LAP. With a pre-assigned threshold of insulin (Table 2, all P < 0.0001). Te associations were sensitivity for detecting a true-responder, greater than still persisted in multivariate linear regression analysis 70% in the present study, 25.5 of the LAP was thought with age and sex adjusted (all P < 0.0001). Compared with to have a more appropriate screening efect for preva- other adiposity measurements, LAP reached the highest lent diabetes, of which the sensitivity and specifcity were 70.0% and 52.0%, respectively. Accordingly, the sensitivity Sun et al. Diabetol Metab Syndr (2019) 11:51 Page 5 of 10

Table 2 Pearson’s correlation and multiple regression analysis of body adiposity indexes associated with glucose metabolism indexes parameters FPG (mmol/L) OGTT 2 h glucose (mmol/L) HbA1c (%) Fasting insulin r P value St. β P value r P value St. β P value r P value St. β P value r P value St. β P value

BMI 0.15 < 0.0001 0.15 < 0.0001 0.17 < 0.0001 0.17 < 0.0001 0.15 < 0.0001 0.15 < 0.0001 0.46 < 0.0001 0.46 < 0.0001 WC 0.19 < 0.0001 0.17 < 0.0001 0.20 < 0.0001 0.18 < 0.0001 0.19 < 0.0001 0.17 < 0.0001 0.45 < 0.0001 0.49 < 0.0001 WHR 0.18 < 0.0001 0.15 < 0.0001 0.20 < 0.0001 0.18 < 0.0001 0.19 < 0.0001 0.17 < 0.0001 0.28 < 0.0001 0.33 < 0.0001 WHtR 0.18 < 0.0001 0.16 < 0.0001 0.23 < 0.0001 0.20 < 0.0001 0.20 < 0.0001 0.17 < 0.0001 0.47 < 0.0001 0.47 < 0.0001 VAI 0.21 < 0.0001 0.21 < 0.0001 0.27 < 0.0001 0.26 < 0.0001 0.22 < 0.0001 0.21 < 0.0001 0.43 < 0.0001 0.42 < 0.0001 BAI 0.05 < 0.0001 0.09 < 0.0001 0.11 < 0.0001 0.13 < 0.0001 0.07 < 0.0001 0.09 < 0.0001 0.32 < 0.0001 0.35 < 0.0001 LAP 0.22 < 0.0001 0.20 < 0.0001 0.27 < 0.0001 0.25 < 0.0001 0.24 < 0.0001 0.22 < 0.0001 0.52 < 0.0001 0.52 < 0.0001

Multiple regression analysis is adjusted for age and sex BMI body mass index, WC waist circumference, WHR waist-hip ratio, WHtR waist-height ratio, VAI visceral adiposity index, BAI body adiposity index, LAP lipid accumulation product index, FPG fasting plasma glucose, OGTT​ oral glucose tolerance test r correlation coefcient, St. β Standardized regression coefcient and specifcity were 81.0% and 59.0% by using threshold Te BAI has emerged as the preferred parameter to 25.5 of the LAP for screening prevalent insulin resistance. estimates body composition as its correlation coef- cient with DXA-derived percentage body fat was as high as 0.85 in Mexican–American and black subjects [10]. Discussion Recently, Alvim et al. [39] proposed that BAI is a better By selecting widely used clinical parameters of body adi- indicator for diabetes than BMI and WC in the Amer- pose composition, this large-scale research in middle- indian population. However, its predictive power for aged and elderly Chinese suggests that elevated LAP has percentage body fat or diabetes in populations of other better relationship with both diabetes and insulin resist- ethnicities seem to be inaccurate and inconsistent [9, ance than other adiposity measures tested in the study. 40]. Te results of these studies, together with our pre- Measurement of BAI, however, showed considerably sent fndings, suggested that BAI does not appear to be a weakest correlation with diabetes and insulin resistance prominent anthropometric indicator of glycometabolism when compared with other body adipose indexes. It is disorder in Chinese population. noteworthy that these conclusions highlight the clinical VAI is a novel indicator to assess both visceral fat distri- importance of considering LAP into the evaluation tar- bution and adipose tissue dysfunction, which is reported get for diabetes control, especially in or obese to be closely correlated with risk of impaired glucose adults. metabolism and diabetes [41, 42]. However, there are also A better understanding of the variation in distinct studies suggested that VAI is unlikely to improve the pre- adiposity measures would probably shed light on the diction ability of type 2 diabetes beyond its components determination of obesity heterogeneity in subjects with such as BMI and WC, and meanwhile, the performance diabetes. Despite BMI and WC are commonly measure- of VAI for the diagnose of diabetes was not better than ments of individual obesity, utility of these parameters commonly available information on WHtR [43, 44]. In in predicting body fat is particularly inaccurate in dis- the present study, compared to other commonly available tinguish between visceral adipose tissue and subcutane- body adiposity indices such as BMI and WC, we found ous adiposity tissue. Te increasing recognition of the that VAI has a better in identifcation ability of prevalent diferences in metabolic profles between fat and fat-free diabetes. In addition, our results show that VAI is better mass have also stimulated interest in exploring simple than WHtR in the risk assessment for prevalent diabe- and efective anthropometric indicators for evaluation of tes. Nevertheless, despite all strongly and independently visceral adiposity. In the Dallas Heart study, WHR more related to glycometabolism parameters, the above meas- accurate in estimating the risk of than ures involving WHtR, BMI and WC, have been shown to BMI and has been reported to be the best measurement be superior to VAI in the association with insulin resist- of body adipose tissue mass [36, 37]. Nevertheless, results ance based on current fndings. of the SAPHIR study stated that WHR was not inferior to As a promising measurement of abdominal adiposity, BAI, BMI, or WHtR in its calculation of anthropometri- LAP is frst developed to refect the combined anatomic cal estimations, and the measurement of glucose homeo- and physiologic changes, which can accurately difer- stasis was hardly evaluated by WHR [38]. entiate between visceral adiposity and subcutaneous Sun et al. Diabetol Metab Syndr (2019) 11:51 Page 6 of 10

Fig. 1 Distribution of FPG, OGTT 2 h glucose and HbA1c stratifed by quartiles of body adiposity parameters a FPG, b OGTT 2 h glucose, c HbA1c. The box plot: upper horizontal line of box, 75th percentile; lower horizontal line of box, 25th percentile; solid point within box, 50th percentile (the median); upper solid point outside box, 90th percentile; lower solid point outside box, 10th percentile

adiposity by describing over accumulation of fat mass procedure. Actually, the process of LAP calculation is [45, 46]. Te superiority of this adiposity assessment simple as (WC − 65) × TG in men and (WC − 58) × TG techniques is that a higher score will indicate a higher in women, which ofers a better applicability in clinical degree of lipid accumulation in the body. By secret- practice. Terefore, LAP could refect visceral adiposity ing adipocytokines and increasing plasma concentra- accumulation and be used as a non-invasive assessment tion of free fatty acid, visceral fat is recently considered technique for assessing insulin resistance and diabetes. to be one of the multifunctional organ, which can inter- Some limitations of the study should be discussed. fere with the insulin signal and lead to insulin resistance First, no causal inference can be drawn due to the cross- and diabetes [47]. It has been found that individuals with sectional design of the current study. Further prospective greater degree of visceral fat also have greater degree of cohort studies are needed to confrm the precise rela- insulin resistance [48]. Besides accuracy, the ideal visceral tionship between LAP and incidence of diabetes. Sec- adiposity parameter should be with simple calculation ond, the imaging techniques such as DXA and magnetic Sun et al. Diabetol Metab Syndr (2019) 11:51 Page 7 of 10

Fig. 2 Prevalence of diabetes and insulin resistance in diferent quartiles of body adiposity parameters a diabetes, b insulin resistance

resonance imaging are more accurate techniques for Conclusion quantifying body fat distribution, but these examinations Our fndings add further evidence to understand the are costly and time consuming to be applied routinely in heterogeneity in the association of body adiposity clinical settings. However, sampling analysis in assessing compartments with prevalence of diabetes and insu- the concordance between generalizability of adiposity lin resistance. LAP may be suggested as an applicable parameters using in this study and those reliable imag- parameter in assessing relationship between excess vis- ing techniques should be conducted to strengthen the ceral fat mass and diabetes. fndings of the present study. Tird, fndings of our study derived insight into the prominent role of LAP than other specifc parameters of adipose tissue distribution in Key messages assessing diabetes and insulin resistance in subjects aged 40 years or older. In clinical settings, however, to improve 1. Te present fndings add further evidence to under- the accuracy of estimating body adiposity, the LAP may stand the heterogeneity in the association of body also need to be frstly optimized to take account of age adiposity compartments with prevalent diabetes and diferences in body composition. Moreover, the study insulin resistance. only including Chinese subjects and the results might 2. Te LAP index may be suggested as an applicable not be representative of other ethnic groups, especially indicator in assessing relationship between excess for those in the developed or undeveloped countries. To visceral fat mass and diabetes. some extent, the population in the present study was still 3. Te BAI does not appear to be a prominent anthro- a convenience sample and selection bias is inevitable. pometric indicator of glycometabolism disorder in Nevertheless, the present study, to our current knowl- Chinese population based on the present fndings. edge, is the largest population-based study to explore the association of routine and non-traditional body adipos- ity indicators with both diabetes and insulin resistance in Chinese. Sun et al. Diabetol Metab Syndr (2019) 11:51 Page 8 of 10

Table 3 Association of increased body adiposity parameters with risk of prevalent diabetes and insulin resistance 1-Quartile change of increasing body adiposity parameters AUC (95% CI) P value Model 1 Model 2 Model 3

Diabetes BMI 1.42 (1.36–1.49) 1.43 (1.37–1.50) 1.41 (1.33–1.49) 0.616 (0.602–0.630) < 0.0001 WC 1.53 (1.46–1.60) 1.46 (1.39–1.54) 1.44 (1.35–1.52) 0.638 (0.624–0.651) 0.0003 WHR 1.57 (1.50–1.65) 1.47 (1.40–1.54) 1.44 (1.36–1.51) 0.648 (0.635–0.661) 0.140 WHtR 1.55 (1.48–1.63) 1.48 (1.41–1.55) 1.45 (1.37–1.54) 0.646 (0.632–0.659) 0.0276 VAI 1.57 (1.50–1.64) 1.54 (1.47–1.61) 1.49 (1.41–1.58) 0.645 (0.632–0.658) 0.0002 BAI 1.18 (1.13–1.24) 1.18 (1.13–1.24) 1.23 (1.15–1.30) 0.555 (0.540–0.569) < 0.0001 LAP 1.65 (1.57–1.73) 1.60 (1.53–1.68) 1.60 (1.50–1.69) 0.658 (0.645–0.671) – Insulin resistance BMI 2.32 (2.20–2.44) 2.32 (2.21–2.44) 2.21 (2.09–2.33) 0.752 (0.741–0.763) < 0.0001 WC 2.27 (2.16–2.39) 2.25 (2.14–2.37) 2.28 (2.15–2.41) 0.738 (0.727–0.749) < 0.0001 WHR 1.74 (1.67–1.82) 1.73 (1.65–1.81) 1.73 (1.64–1.82) 0.678 (0.666–0.690) < 0.0001 WHtR 2.26 (2.15–2.37) 2.24 (2.13–2.36) 2.11 (2.00–2.22) 0.744 (0.733–0.755) < 0.0001 VAI 2.21 (2.11–2.32) 2.20 (2.09–2.31) 2.04 (1.93–2.15) 0.738 (0.726–0.749) < 0.0001 BAI 1.58 (1.51–1.65) 1.58 (1.51–1.65) 1.65 (1.56–1.74) 0.648 (0.635–0.661) < 0.0001 LAP 2.70 (2.56–2.85) 2.69 (2.55–2.84) 2.54 (2.40–2.69) 0.781 (0.771–0.792) –

Data are odds ratios (95% confdence interval). Participants without diabetes or insulin resistance are defned as 0 and with diabetes or insulin resistance as 1 P value is calculated by the nonparametric approach to compare AUC of other anthropometric measures with the LAP group Model 1 is unadjusted Model 2 is adjusted for age Model 3 is adjusted for age, sex, current smoking and drinking status, physical activity level, SBP, LDL-C, γ-GGT, eGFR and antidiabetic treatment BMI body mass index, WC waist circumference, WHR waist-hip ratio, WHtR waist-height ratio, VAI visceral adiposity index, BAI body adiposity index, LAP lipid accumulation product index, AUC​ area under the receiver operating characteristic curve

Table 4 The risk of prevalent diabetes and insulin resistance according to quartiles of increased body adiposity parameters Quartile 1 Quartile 2 Quartile 3 Quartile 4

Diabetes BMI 1 1.58 (1.30–1.92) 1.96 (1.63–2.36) 2.92 (2.43–3.50) WC 1 1.33 (1.09–1.63) 1.85 (1.53–2.24) 2.89 (2.39–3.50) WHR 1 1.58 (1.30–1.93) 2.16 (1.77–2.63) 3.10 (2.55–3.77) WHtR 1 1.39 (1.14–1.71) 1.80 (1.50–2.16) 3.05 (2.55–3.64) VAI 1 1.50 (1.23–1.82) 2.13 (1.76–2.58) 3.34 (2.78–4.01) BAI 1 1.25 (1.05–1.50) 1.37 (1.14–1.66) 1.88 (1.56–2.27) LAP 1 1.58 (1.28–1.94) 2.27 (1.86–2.77) 4.06 (3.34–4.94) Insulin resistance BMI 1 2.84 (2.31–3.49) 4.88 (4.01–5.95) 12.19 (10.04–14.79) WC 1 2.72 (2.20–3.38) 5.11 (4.17–6.28) 12.96 (10.55–15.93) WHR 1 2.11 (1.77–2.52) 3.68 (3.08–4.38) 5.58 (4.68–6.67) WHtR 1 2.44 (1.99–3.00) 4.24 (3.52–5.11) 9.97 (8.29–11.99) VAI 1 2.30 (1.89–2.80) 4.37 (3.62–5.26) 8.98 (7.48–10.79) BAI 1 1.92 (1.62–2.27) 2.84 (2.39–3.38) 4.74 (3.98–5.66) LAP 1 3.09 (2.45–3.90) 6.62 (5.30–8.27) 18.27 (14.66–22.77)

Data are odds ratios (95% confdence interval). Participants without diabetes or insulin resistance are defned as 0 and with diabetes or insulin resistance as 1 All models are adjusted for age, sex, current smoking and drinking status, physical activity level, SBP, LDL-C, γ-GGT, eGFR and antidiabetic treatment BMI body mass index, WC waist circumference, WHR waist-hip ratio, WHtR waist-height ratio, VAI visceral adiposity index, BAI body adiposity index, LAP lipid accumulation product index Sun et al. Diabetol Metab Syndr (2019) 11:51 Page 9 of 10

Abbreviations 4. Cornier MA, Despres JP, Davis N, Grossniklaus DA, Klein S, Lamarche SBP: systolic blood pressure; DBP: diastolic blood pressure; TG: triglycerides; TC: B, et al. Assessing adiposity: a scientifc statement from the American total cholesterol; HDL-C: high-density lipoprotein cholesterol; LDL-C: low-den- Heart Association. Circulation. 2011;124:1996–2019. sity lipoprotein cholesterol; FPG: fasting plasma glucose; OGTT​: oral glucose 5. Du T, Yuan G, Zhang M, Zhou X, Sun X, Yu X. Clinical usefulness of tolerance test; HOMA-IR: homeostasis model assessment of insulin resistance; lipid ratios, visceral adiposity indicators, and the triglycerides and eGFR: estimated glomerular fltration rate; γ-GGT​: γ-glutamyltransferase; BMI: glucose index as risk markers of insulin resistance. Cardiovasc Diabetol. body mass index; WC: waist circumference; WHR: waist-hip ratio; WHtR: waist- 2014;13:146. height ratio; VAI: visceral adiposity index; BAI: body adiposity index; LAP: lipid 6. Nevill AM, Metsios GS. The need to redefne age- and gender-specifc accumulation product index; MET-h/wk: separate metabolic equivalent hours overweight and obese body mass index cutof points. Nutr Diab. per week. 2015;5:e186. 7. Rothman KJ. BMI-related errors in the measurement of obesity. Int J Acknowledgements Obes. 2008;32(Suppl 3):S56–9. We are indebted to the participants in the present study for their persis- 8. Nevill AM, Stavropoulos-Kalinoglou A, Metsios GS, Koutedakis Y, Holder tent outstanding support and to our colleagues working for their valuable RL, Kitas GD, et al. Inverted BMI rather than BMI is a better proxy for assistance. percentage of body fat. Ann Hum Biol. 2011;38:681–4. 9. Schulze MB, Thorand B, Fritsche A, Haring HU, Schick F, Zierer A, et al. Authors’ contributions Body adiposity index, body fat content and incidence of type 2 diabe- Conceived and designed the experiments: YL and KS. Performed the experi- tes. Diabetologia. 2012;55:1660–7. ments: FL, YQ, WF, CC, KS and DL. Analyzed the data: KS, QF and MR Wrote the 10. Bergman RN, Stefanovski D, Buchanan TA, Sumner AE, Reynolds manuscript: KS and DL. All authors read and approved the fnal manuscript. JC, Sebring NG, et al. A better index of body adiposity. Obesity. 2011;19:1083–9. Funding 11. Freedman DS, Thornton JC, Pi-Sunyer FX, Heymsfeld SB, Wang J, This work was supported by Grants from: (1) The National Natural Science Pierson RN Jr, et al. The body adiposity index (hip circumference/ Foundation of China (81600642, 81471034, 81370910, 81300675); (2) Natural height(1.5)) is not a more accurate measure of adiposity than is BMI, Science Foundation of Guangdong Province, China (2015A030310433, waist circumference, or hip circumference. Obesity. 2012;20:2438–44. 2017A030313831); (3) Sun Yat-Sen University Medical 2016 Youth Teacher 12. Sun G, Cahill F, Gulliver W, Yi Y, Xie Y, Bridger T, et al. Concordance of Research Funding Project (16ykpy27); (4) Sun Yat-Sen Clinical Research Culti- BAI and BMI with DXA in the Newfoundland population. Obesity. vating Program (SYS-Q-201801); (5) Major project of the people’s livelihood 2013;21:499–503. science and technology in Guangzhou (201300000102); (6) The 863 project 13. Amato MC, Giordano C, Galia M, Criscimanna A, Vitabile S, Midiri M, of Young Scientist (SS2015AA020927); (7) The Zhujiang Star of science and et al. Visceral Adiposity Index: a reliable indicator of visceral fat function technology Foundation in Guang Zhou (2014J2200046); (8) Grants from the associated with cardiometabolic risk. Diabetes Care. 2010;33:920–2. Chinese Society of Endocrinology and National Clinical Research Center for 14. Ji B, Qu H, Wang H, Wei H, Deng H. Association between the visceral Metabolic Diseases; (9) State Key Clinical Specialty Construction Project (2011); adiposity index and homeostatic model assessment of insulin resist- (10) Science and Technology Planning Project of Guangdong Province, China ance in participants with normal waist circumference. Angiology. (2014A020212161). The funders had no role in study design, data collection 2017;68:716–21. and analysis, decision to publish, or preparation of the manuscript. 15. Wu J, Gong L, Li Q, Hu J, Zhang S, Wang Y, et al. A novel visceral adipos- ity index for prediction of type 2 diabetes and pre-diabetes in Chinese Availability of data and materials adults: a 5-year prospective study. Sci Rep. 2017;7:13784. The work described was original research that has not been published 16. Liu C, Feng X, Li Q, Wang Y, Li Q, Hua M. Adiponectin, TNF-alpha and previously, and not under consideration for publication elsewhere, in part infammatory cytokines and risk of type 2 diabetes: a systematic review or in whole. All authors believe that the manuscript represents valid work and meta-analysis. Cytokine. 2016;86:100–9. and have reviewed and approved the fnal version. Main document data and 17. Dutra MT, Gadelha AB, Nobrega OT, Lima RM. Body adiposity index, but additional unpublished data from the study are available by sending Email to not visceral adiposity index, correlates with infammatory markers in [email protected] with proper purposes. sarcopenic obese elderly women. Exp Aging Res. 2017;43:291–304. 18. Wakabayashi I, Daimon T. A strong association between lipid accumu- Ethics approval and consent to participate lation product and diabetes mellitus in Japanese women and men. J All procedures performed in studies involving human participants were in Atheroscler Thromb. 2014;21:282–8. accordance with the ethical standards of the institutional and/or national 19. Hosseinpanah F, Barzin M, Erfani H, Serahati S, Ramezani Tehrani F, Azizi research committee and with the 1964 Helsinki declaration and its later F. Lipid accumulation product and insulin resistance in Iranian PCOS amendments or comparable ethical standards. prevalence study. Clin Endocrinol. 2014;81:52–7. 20. Ma CM, Lu N, Wang R, Liu XL, Lu Q, Yin FZ. Three novel obese indicators Competing interests perform better in monitoring management of metabolic syndrome in The authors declare that they have no competing interests. type 2 diabetes. Sci Rep. 2017;7:9843. 21. Mirmiran P, Bahadoran Z, Azizi F. Lipid accumulation product is associ- Received: 23 April 2019 Accepted: 24 June 2019 ated with insulin resistance, lipid peroxidation, and systemic infamma- tion in type 2 diabetic patients. Endocrinol Metab. 2014;29:443–9. 22. Bi Y, Lu J, Wang W, Mu Y, Zhao J, Liu C, et al. Cohort profle: risk evalua- tion of cancers in Chinese diabetic individuals: a longitudinal (REAC- TION) study. J Diabetes. 2014;6:147–57. 23. Ning G. Risk Evaluation of cAncers in Chinese diabeTic Individuals: a References lONgitudinal (REACTION) study. J Diabetes. 2012;4:172–3. 1. Bragg F, Holmes MV, Iona A, Guo Y, Du H, Chen Y, et al. Association 24. Sun K, Li F, Qi Y, Lin D, Ren M, Xu M, et al. Sex diference in the associa- between diabetes and cause-specifc mortality in rural and urban areas tion between habitual daytime napping and prevalence of diabetes: a of China. JAMA. 2017;317:280–9. population-based study. Endocrine. 2016;52:263–70. 2. Emerging Risk Factors C, Sarwar N, Gao P, Seshasai SR, Gobin R, 25. Sun K, Liu J, Ning G. Active smoking and risk of metabolic syndrome: a Kaptoge S, et al. Diabetes mellitus, fasting blood glucose concentra- meta-analysis of prospective studies. PLoS ONE. 2012;7:e47791. tion, and risk of vascular disease: a collaborative meta-analysis of 102 26. Tomioka K, Iwamoto J, Saeki K, Okamoto N. Reliability and validity prospective studies. Lancet. 2010;375:2215–22. of the International Physical Activity Questionnaire (IPAQ) in elderly 3. Tobias DK, Pan A, Jackson CL, O’Reilly EJ, Ding EL, Willett WC, et al. adults: the Fujiwara-kyo Study. J Epidemiol. 2011;21:459–65. Body-mass index and mortality among adults with incident type 2 diabetes. N Engl J Med. 2014;370:233–44. Sun et al. Diabetol Metab Syndr (2019) 11:51 Page 10 of 10

27. Ma YC, Zuo L, Chen JH, Luo Q, Yu XQ, Li Y, et al. Modifed glomerular 39. Alvim Rde O, Mourao-Junior CA, de Oliveira CM, Krieger JE, Mill JG, Pereira fltration rate estimating equation for Chinese patients with chronic AC. Body mass index, waist circumference, body adiposity index, and risk kidney disease. J Am Soc Nephrol. 2006;17:2937–44. for type 2 diabetes in two populations in Brazil: general and Amerindian. 28. Amato MC, Giordano C. Visceral adiposity index: an indicator of adi- PLoS ONE. 2014;9:e100223. pose tissue dysfunction. Int J Endocrinol. 2014;2014:730827. 40. Ramirez-Velez R, Correa-Bautista JE, Gonzalez-Ruiz K, Vivas A, Triana-Reina 29. Tian S, Xu Y, Han F. Higher predictability of the lipid accumulation HR, Martinez-Torres J, et al. Body adiposity index performance in estimat- product than commonly used anthropometric parameters partly due ing in colombian college students: fndings from the to its defnition. Clin Nutr. 2017;36:909. FUPRECOL-adults study. Nutrients. 2017;9:40. 30. Nascimento-Ferreira MV, Rendo-Urteaga T, Vilanova-Campelo RC, 41. Chen C, Xu Y, Guo ZR, Yang J, Wu M, Hu XS. The application of visceral Carvalho HB, da Paz Oliveira G, Paes Landim MB, et al. The lipid accu- adiposity index in identifying type 2 diabetes risks based on a prospec- mulation product is a powerful tool to predict metabolic syndrome in tive cohort in China. Lipids Health Dis. 2014;13:108. undiagnosed Brazilian adults. Clin Nutr. 2017;36:1693–700. 42. Al-Daghri NM, Al-Attas OS, Alokail MS, Alkharfy KM, Charalampidis P, Liva- 31. Levy JC, Matthews DR, Hermans MP. Correct homeostasis model das S, et al. Visceral adiposity index is highly associated with adiponectin assessment (HOMA) evaluation uses the computer program. Diabetes values and glycaemic disturbances. Eur J Clin Invest. 2013;43:183–9. Care. 1998;21:2191–2. 43. Bozorgmanesh M, Hadaegh F, Azizi F. Predictive performance of the vis- 32. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner ceral adiposity index for a visceral adiposity-related risk: type 2 diabetes. RC. Homeostasis model assessment: insulin resistance and beta-cell Lipids Health Dis. 2011;10:88. function from fasting plasma glucose and insulin concentrations in 44. Wang Y, He S, He J, Wang S, Liu K, Chen X. Predictive value of visceral man. Diabetologia. 1985;28:412–9. adiposity index for type 2 diabetes mellitus: a 15-year prospective cohort 33. American Diabetes A. Classifcation and diagnosis of diabetes: stand- study. Herz. 2015;40(Suppl 3):277–81. ards of medical care in diabetes-2018. Diabetes Care. 2018;41:S13–27. 45. Kahn HS. The “lipid accumulation product” performs better than the 34. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under body mass index for recognizing cardiovascular risk: a population-based two or more correlated receiver operating characteristic curves: a non- comparison. BMC Cardiovasc Disord. 2005;5:26. parametric approach. Biometrics. 1988;44:837–45. 46. Roriz AK, Passos LC, de Oliveira CC, Eickemberg M, Moreira Pde A, Sam- 35. Rj G, Cc P, Lr M, Xm T. Determination and interpretation of the optimal paio LR. Evaluation of the accuracy of anthropometric clinical indicators operating point for ROC curves derived through generalized linear mod- of visceral fat in adults and elderly. PLoS ONE. 2014;9:e103499. els. Understand Stat. 2003;2:219–42. 47. Katsiki N, Mantzoros C, Mikhailidis DP. Adiponectin, lipids and atheroscle- 36. See R, Abdullah SM, McGuire DK, Khera A, Patel MJ, Lindsey JB, et al. rosis. Curr Opin Lipidol. 2017;28:347–54. The association of difering measures of overweight and obesity with 48. Sam S, Hafner S, Davidson MH, D’Agostino RB Sr, Feinstein S, Kondos G, prevalent atherosclerosis: the Dallas Heart Study. J Am Coll Cardiol. et al. Hypertriglyceridemic waist phenotype predicts increased visceral 2007;50:752–9. fat in subjects with type 2 diabetes. Diabetes Care. 2009;32:1916–20. 37. Canoy D. Distribution of body fat and risk of coronary heart disease in men and women. Curr Opin Cardiol. 2008;23:591–8. 38. Melmer A, Lamina C, Tschoner A, Ress C, Kaser S, Laimer M, et al. Body adi- Publisher’s Note posity index and other indexes of body composition in the SAPHIR study: Springer Nature remains neutral with regard to jurisdictional claims in pub- association with cardiovascular risk factors. Obesity. 2013;21:775–81. lished maps and institutional afliations.

Ready to submit your research ? Choose BMC and benefit from:

• fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations • maximum visibility for your research: over 100M website views per year

At BMC, research is always in progress.

Learn more biomedcentral.com/submissions